Yahboom ROSMASTER M3 Pro ROS2 Robot with OpenClaw AI Agent, Dual TOF LiDAR, 6DOF Arm, Mecanum SLAM
Yahboom ROSMASTER M3 Pro ROS2 Robot with OpenClaw AI Agent, Dual TOF LiDAR, 6DOF Arm, Mecanum SLAM
Yahboom
Nie można załadować gotowości do odbioru
Overview
ROSMASTER M3 Pro is a ROS2 Robot platform by Yahboom for ROS education, scientific research experiments, and AI application teaching. It uses a Mecanum wheel chassis with pendulum suspension for omnidirectional movement and is developed on ROS2 Humble. The platform integrates a 6DOF robotic arm, a binocular structured-light depth camera for 3D vision hand-eye integration, and dual TOF LiDAR for omnidirectional SLAM mapping, autonomous navigation, obstacle avoidance, and path planning. It also supports multimodal AI large-model interaction (text/image/voice) with speech recognition and natural language understanding for task planning and execution.
Key Features
- OpenClaw AI agent deployment (with deployment and usage tutorial). Note: OpenClaw deployment is not supported on the Jetson Nano B01 version.
- Embedded multimodal large model capabilities: extensible RAG knowledge base, visual large language model, text large language model, dual-model reasoning architecture, and dynamic feedback reasoning.
- Dual TOF LiDAR point cloud fusion: 360° omnidirectional perception without blind spots; mapping navigation/road network planning; path planning and multi-point navigation.
- Road network planning: create, edit, and manage route networks composed of points and connecting lines; supports shortest-path selection in sandbox-style route networks.
- 6DOF 3D visual robotic arm: 3D space grasping, sorting and transportation; 3D point cloud recognition; target positioning and tracking; distance/volume calculation; 3D real-scene mapping.
- Deep vision technology applications: YOLOv26 / Transformer, MediaPipe / OpenCV, visual fusion repositioning navigation, PCL real-time point cloud segmentation.
- Built-in AI large model voice module and speaker: supports real-time conversion between voice and text.
- MoveIt2 simulation support.
Specifications
| Model | ROSMASTER M3 Pro |
| System | ROS2 Humble |
| Chassis | All-aluminum alloy body; Mecanum wheel pendulum suspension; rear-wheel pendulum suspension structure |
| Wheel size | 80mm Mecanum wheels |
| LiDAR | Dual TOF LiDAR (diagonal offset layout: right front + left rear); 360° scanning |
| LiDAR detection (from comparison chart) | 360° omnidirectional perception; 24m detection distance |
| Depth camera | Binocular structured-light depth camera |
| Depth camera FOV (from comparison chart) | H91° V62° |
| Robotic arm | 6DOF robotic arm; 6PCS intelligent serial bus servos (supports reading back position/status and other information) |
| Gripper capability (from arm description) | Clamps up to 410g; repeatable positioning accuracy 0.5mm |
| Battery | 9600mAh high-capacity battery pack |
| Touch screen | 7-inch IPS high-definition touch screen (optional); configuration variants shown: with display / without display |
| Motors | High torque encoder metal motor; independent swing suspension with high torque motor |
| ROS control board | 3rd generation ROS control board |
| MoveIt | MoveIt2 |
| AI large-model application schemes | OpenClaw AI agent; optional Dify workflow platform |
| OpenClaw AI agent – supported master control | Raspberry Pi 5; Jetson Orin Nano SUPER; Jetson Orin NX SUPER |
| OpenClaw AI agent – interaction methods | Voice, WAP, web/terminal text commands |
| OpenClaw AI agent – robot control mode | MCP, CLI |
| Dify workflow platform – supported master control | Raspberry Pi 5; Jetson Orin Nano SUPER; Jetson Orin NX SUPER; Jetson Nano B01 |
| Dify workflow platform – robot control mode | http |
| AI visual tracking algorithm (from solution comparison) | OpenClaw: Transformer model; Dify: KCF |
| Optional AI large-model scenario sand table / sandbox map | Size: 3m × 4.1m (optional accessory; not included with ROSMASTER M3 Pro) |
Master Control Board Options (for selection)
| Option | Key compute spec shown | Power (shown) | ROS system (shown) | OpenClaw (shown) |
| Jetson Nano B01 4GB | 0.5 TFLOPS (FP16); Quad-Core Arm Cortex-A57 MPCore; 128-core NVIDIA Maxwell GPU; 4GB 64-bit LPDDR4 (25.6 GB/s) | 5W, 10W | Ubuntu 18.04 LTS + Docker + ROS2 Humble | Not supported |
| Raspberry Pi 5 (8GB/16GB) | Cortex-A76; VideoCore VII; RAM: 8GB/16GB | 10W | Raspberry Pi OS + Docker + ROS2 Humble | (See OpenClaw support note above) |
| Jetson Orin Nano SUPER 8GB | 67 TOPS; 6-core Arm Cortex-A78AE v8.2 64-bit CPU (1.5MB L2 + 4MB L3); 1024-core NVIDIA Ampere GPU with 32 Tensor Cores; 8GB 128-bit LPDDR5 (102 GB/s) | 7W, 15W, 25W | Ubuntu 22.04 LTS + ROS2 Humble | Support |
| Jetson Orin NX SUPER 8GB | 117 TOPS; 6-core NVIDIA Arm Cortex-A78AE v8.2 64-bit CPU (1.5MB L2 + 4MB L3); 1024-core NVIDIA Ampere GPU with 32 Tensor Cores; 8GB 128-bit LPDDR5 (102 GB/s) | 10W, 15W, 25W, 40W | Ubuntu 22.04 LTS + ROS2 Humble | Support |
| Jetson Orin NX SUPER 16GB | 157 TOPS; 8-core NVIDIA Arm Cortex-A78AE v8.2 64-bit CPU (2MB L2 + 4MB L3); 1024-core NVIDIA Ampere GPU with 32 Tensor Cores; 16GB 128-bit LPDDR5 (102 GB/s) | 10W, 15W, 25W, 40W | Ubuntu 22.04 LTS + ROS2 Humble | Support |
Functional Case Test Comparison (shown)
| Version | Offline speech recognition / speech synthesis | AI large model task decision planning time | Simple task loading time | Complex task loading time | Tracking & color block grabbing | Advanced 3D visual functions | MediaPipe development | MoveIt2 simulation |
| Raspberry Pi 5 16GB | None | 2s | 10s | 15s | 15fps | 15fps | 15fps | Using a companion virtual machine |
| Jetson Nano B01 4GB | None | 2s | 12s | 13s | 15fps | 15fps | 10fps | Using a companion virtual machine |
| Jetson Orin Nano SUPER 8GB | 4s | 2s | 6s | 8s | 30fps | 30fps | 30fps | 30fps+ |
| Jetson Orin NX SUPER 16GB | 4s | 2s | 4s | 4s | 30fps | 30fps | 30fps | 30fps+ |
For configuration selection help (Raspberry Pi vs Jetson options) or after-sales support, contact https://rcdrone.top/ or email support@rcdrone.top.
Applications
- ROS2 education and labs: SLAM mapping, navigation, obstacle avoidance, and road network planning.
- 3D vision & manipulation: 3D recognition/grasping, sorting, tracking, and handling with a 6DOF arm and depth point cloud.
- Multimodal AI interaction: voice/text/image interaction with task decomposition, long-term scheduling, memory search, and proactive response logic (OpenClaw workflow).
- AI visual recognition (examples shown): human feature recognition, gesture recognition, finger tip trajectory recognition, human skeleton recognition, 3D detection, 3D face detection, tag code recognition, zero-shot Transformer object tracking, visual re-localization fusion navigation solution, rotating object detection and grasping.
- Depth camera functions (examples shown): depth image/point cloud, distance measurement, PCL real-time point cloud segmentation and localization, RTAB-Map 3D visual mapping navigation, regional target height measurement, wood block volume measurement.
- LiDAR functions (examples shown): Gmapping/Cartographer/slam_toolbox mapping, dual LiDAR fusion filtering, DWA dynamic obstacle avoidance, single/multi-point navigation, app mapping navigation, repositioning mapping navigation, road network planning, LiDAR obstacle avoidance, LiDAR following, LiDAR guard.
Manuals
- Tutorial/Study page: https://www.yahboom.net/study/ROSMASTER-M3PRO
Details
An all-in-one ROS2 Humble education platform combining omnidirectional mobility, 3D vision, and a 6DOF robotic arm.
Multimodal interaction and autonomy features support mapping, navigation, grasping, and task execution in one platform.
OpenClaw enables natural-language task planning with options for voice, app, and text-based commands.
Dual TOF LiDAR fusion delivers 360° perception for SLAM mapping, obstacle avoidance, and flexible route planning.
Three built-in model types cover text reasoning, voice interaction, and visual understanding for richer robotics demos.
Choose between embedded OpenClaw deployment or an optional workflow platform depending on your project needs.
A modular scenario table supports repeatable training scenes for sorting, counting, and navigation exercises.
Example projects highlight how agent-based control can be applied to everyday lab tasks and interactive demos.
Agent workflows can connect chat-based instructions with mapping, navigation, and transport behaviors.
Tools like memory search and MCP-style calling help connect higher-level intent to reliable robot actions.
Vision-driven behaviors include target tracking, color recognition, autonomous cruising, and coordinated arm actions.
Binocular structured-light depth sensing supports hand–eye coordination for 3D measurement, recognition, and grasping.
Configuration comparisons help select the right sensing and compute combination for your classroom or lab.
A selection guide summarizes common configurations and differences across function sets.
Core ROS capabilities cover LiDAR mapping, depth-camera perception, and visual recognition pipelines.
MoveIt2 simulation and motion-control demos support planning, grasping workflows, and multi-robot coordination.
An aluminum mecanum chassis with pendulum suspension improves stability while keeping full ROS2 Humble compatibility.
Multiple control methods and a clear structure layout make it easier to set up, maintain, and expand the robot.
The ROSMASTER M3 Pro platform can be configured with a 6DOF robotic arm and a binocular structured light depth camera for grasping and depth-based perception tasks.
TOF laser LiDAR supports 0.05–12 m ranging with up to 4000 scans per second, while the voice module adds mic and speaker connections for voice interaction.
The Yahboom ROSMASTER M3 Pro ROS2 robot control board provides a compact, labeled connector layout for building and expanding a mobile robot system.
Yahboom ROSMASTER M3 Pro includes access to 200+ detailed courses via an online tutorial repository for learning ROS2 and AI.
The ROSMASTER M3 Pro learning outline covers ROS control basics alongside OpenCV vision tasks, SLAM mapping, and AI features for progressive ROS2 practice.
The ROSMASTER M3 Pro learning roadmap covers topics like OpenCV vision, MediaPipe tracking, MoveIt2 simulation, and ROS2 basics.
The ROSMASTER M3 Pro includes open-source code folders and detailed tutorials covering ROS basics, mapping, navigation, and vision tasks.
ROSMASTER M3 Pro comes with ROS2 video tutorials with English subtitles and provides 3D model files to support development and integration.
The ROSMASTER Series comparison outlines key differences in chassis type, RGBD camera options, control boards, and battery capacity to help choose the right ROS2 robot platform.
The ROSMASTER M1 platform pairs a mecanum-wheel chassis and 520 geared motors with selectable camera, LiDAR, and control-board options for ROS-based development.
ROSMaster M3 Pro pairs a mecanum-wheel chassis with RGBD camera options, 0.91-inch OLED/optional 7-inch touchscreen, and a 12.6V 6000mAh battery.
The ROSMASTER M3 Pro platform lists a mecanum-wheel chassis, optional RGBD camera, 6‑DOF robotic arm, dual LiDAR, and Raspberry Pi or Jetson control board options.
The Yahboom ROSMASTER M3 Pro ROS2 robot’s dimension drawings list key overall measurements in millimeters for planning fit and mounting.
ROSMASTER M3 Pro supports Raspberry Pi 5 or Jetson Orin platforms with Python programming, WiFi networking, and a 12.6V 9600mAh battery pack.
The ROSMASTER M3 Pro kit includes the robot chassis, 6DOF arm, controller, expansion boards, batteries, brackets, and basic tools for assembly.
Optional accessory bundles are organized by controller option, including a 7-inch touch screen set and kits for Raspberry Pi or NVIDIA Jetson boards with the needed cables and mounts.
Related Collections
